Glioblastoma (GBM) remains among the most formidable cancers to treat, and current therapies, including radiation and temozolomide chemotherapy, have only palliative effects. Tumor recurrence occurs in virtually all patients, and relapsed tumors are more resistant to current cytotoxic therapies. Recent robust transcriptional profiling has classified GBM into 4 subtypes with distinct clinical features. In particular, mesenchymal (MES) GBM appears to be the most common subtype with the poorest prognosis. Nonetheless, the mechanisms that regulate growth of MES GBM cells have not been clarified. Among the heterogeneous cells in tumors, glioma stem cells (GSCs) are one of the, if not the only, critical therapeutic targets; however, subtype-specific GSCs are poorly characterized. Our preliminary data suggest the following: a- mesenchymal, but not proneural, GSCs express CD44v6; b- siRNA for CD44v6 reduces in vitro growth of MES, but not PN GSCs; c- CD44v6(+) cells but not CD44v6 (-) cells in mesenchymal GSCs express the neural stem cell-associated gene MELK; d- siRNA for CD44v6 reduces MELK expression; e- CD44 locus exhibits multiple Musashi 1 (Msi1)-binding sequences; f- shRNA- mediated Msi1 depletion diminishes CD44 expression and negatively affects splicing of exon v6. To achieve the goals of this study, we established and characterized patient-derived GBM sphere cultures and created patient-derived mouse GBM tumor models that recapitulate the histopathology of the original tumors. With these pre-clinical models, we will test the hypothesis that the Msi1-CD44v6 signaling is required for the growth and survival of therapy-resistant mesenchymal glioblastoma. Specifically, in Aim 1, we will test the hypothesis that CD44v6 is functionally essential for the proliferation of MES GSCs through signal interaction with MELK. In Aim 2, we will determine whether CD44 transcripts are a direct target of the RNA-binding protein Msi1 in GBM and if Msi1 via the axis CD44v6-MELK plays a role in radio-resistance.